Re: Does SOA expire for root matters?
Brandon Martin via NANOG <[email protected]>
| Newsgroups | gmane.org.operators.nanog |
|---|---|
| Message-ID | <[email protected]> |
On 5/8/26 09:44, Anurag Bhatia via NANOG wrote: > o 604800 (1 week) would actually make the DNS replicas to stop resolving > if disconnected for an extended period of over a week from their master? Unless the server stores a negative/expiry result for the domain, which I don't think any major implementations do, it should just fall back to its hints file which would allow it to bootstrap itself again the same as upon initial startup. In that respect, I don't think the SOA expiry is a big deal. You want it to be long enough that normally-running systems won't constantly keep having to resolve it or fall back to their hints, but you also don't want it super long so that they won't pick up actual changes (which allow long-lived servers to keep running even if their hints file is out of date). A week seems reasonable in both respects. I assume that replication of the root zone among the actual root servers is handled by means other than simple AXFR/IXFR. -- Brandon Martin _______________________________________________ NANOG mailing list https://lists.nanog.org/archives/list/[email protected]/message/VQKNQ77RPGQCVY4IYOVCHE55V5LZJXGA/